arXiv subjects

Carlos Patiño

Publications and source records attributed to Carlos Patiño.

2 recordsLinked to original sources

arXiv2025

Theoretical Characterization of the Magnetic Properties of Vanadium-doped Ti2C MXenes

MXenes are two-dimensional materials composed of transition metals and light elements, known for their high conductivity and versatile surface chemistry. The introduction of spin centers via doping can lead to promising materials for spintronics, magnetic sensing, and data storage. We study the effect of V doping in Ti2C using first principle and Monte-Carlo simulations. Our results show that pristine Ti2C exhibits ferromagnetic intralayer and antiferromagnetic interlayer exchange interactions, yielding an antiferromagnetic ground state. Vanadium incorporation alters these couplings, yet all doped configurations - Ti7VC4 and three Ti6V2C4 variants - retain predominantly antiferromagnetic order. Despite the preserved ground state, V doping enhances the magnetic response, most notably in the p-Ti6V2C4 configuration, which displays increased low-field susceptibility and partial spin alignment across layers. As experimentally isolating individual doped phases is unlikely, samples will consist of mixed configurations whose collective behavior nonetheless exhibits clear signatures of V-induced magnetic modification. These results reveal how transition-metal substitution tunes exchange interactions in MXenes and offer guidance for engineering their magnetic functionalities.

cond-mat.mtrl-sci↗

arXiv2025

Utility-Directed Conformal Prediction: A Decision-Aware Framework for Actionable Uncertainty Quantification

Interest has been growing in decision-focused machine learning methods which train models to account for how their predictions are used in downstream optimization problems. Doing so can often improve performance on subsequent decision problems. However, current methods for uncertainty quantification do not incorporate any information about downstream decisions. We develop a methodology based on conformal prediction to identify prediction sets that account for a downstream cost function, making them more appropriate to inform high-stakes decision-making. Our approach harnesses the strengths of conformal methods -- modularity, model-agnosticism, and statistical coverage guarantees -- while incorporating downstream decisions and user-specified utility functions. We prove that our methods retain standard coverage guarantees. Empirical evaluation across a range of datasets and utility metrics demonstrates that our methods achieve significantly lower costs than standard conformal methods. We present a real-world use case in healthcare diagnosis, where our method effectively incorporates the hierarchical structure of dermatological diseases. The method successfully generates sets with coherent diagnostic meaning, potentially aiding triage for dermatology diagnosis and illustrating how our method can ground high-stakes decision-making employing domain knowledge.

cs.LG↗